Skip to content

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Douyin Transcriber (dyt)

A CLI tool that transcribes Douyin (抖音) video speech to text.

Features

  • 🔗 Smart URL resolution — supports short links (v.douyin.com), full URLs, and pasted share text
  • 📥 Direct CDN download — no yt-dlp needed, no cookies, no login required
  • 🎙️ Dual transcription modes:
    • Local mode — whisper.cpp (free, private, offline)
    • Cloud mode — OpenAI Whisper API (faster, more accurate)
  • 🌐 Multi-language — supports Chinese, English, and all languages whisper supports

Installation

brew install vangie/formula/douyin-transcriber

Prerequisites

  • ffmpegbrew install ffmpeg (audio extraction)
  • Local modebrew install whisper-cpp (auto-downloads models on first run)
  • Cloud modeexport OPENAI_API_KEY=sk-...

Usage

# Paste a Douyin share link — local mode (auto-detected when no API key)
dyt "https://v.douyin.com/n234HvHgzN0/"

# Full URL works too
dyt "https://www.douyin.com/video/7616660340146752819"

# Paste entire share text — URL is auto-extracted
dyt "8.46 复制打开抖音,看看【理性避难所的作品】第9集... https://v.douyin.com/n234HvHgzN0/ 05/31"

# Force local mode with language hint
dyt --local -l zh "https://v.douyin.com/abc123/"

# Cloud mode with custom endpoint (e.g., compatible API)
dyt --cloud --endpoint "https://api.example.com/v1" "https://v.douyin.com/abc123/"

# Save to file
dyt -o transcript.txt "https://v.douyin.com/abc123/"

# Keep the downloaded audio
dyt --keep-audio "https://v.douyin.com/abc123/"

Mode Auto-Detection

Condition Mode
OPENAI_API_KEY set Cloud (OpenAI Whisper API)
OPENAI_API_KEY not set Local (whisper.cpp)
--local flag Force local
--cloud flag Force cloud

Local Mode Options

# Use a different model (tiny/base/small/medium/large-v3)
dyt --model-name small -l zh "URL"

# Use a specific model file
dyt --model-path /path/to/ggml-large-v3.bin "URL"

# Control thread count
dyt -t 8 "URL"

How It Works

  1. Resolve — short URL → follow redirects → extract video ID
  2. Fetch — get video info from Douyin's mobile web page (no API key/cookies needed)
  3. Download — grab audio from CDN directly
  4. Transcribe — whisper.cpp (local) or OpenAI API (cloud)

License

MIT

About

CLI tool to transcribe Douyin video speech to text

Resources

Code of conduct

Contributing

Security policy

Stars

11 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages